A feasibility study of scaling-up the electrolytic production of carbon nanotubes in molten salts

 Real Estate

of 12
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
A feasibility study of scaling-up the electrolytic production of carbon nanotubes in molten salts
  A feasibility study of scaling-up the electrolytic production of carbonnanotubes in molten salts Aleksandar T. Dimitro v  1 , George Z. Chen*, Ian A. Kinloch, Derek J. Fray* Department of Material Science and Metallurgy, Uni  v ersity of Cambridge, Pembroke Street, Cambridge CB2 3QZ, UK  Recei v ed 18 July 2002; recei v ed in re v ised form 28 August 2002 Abstract The feasibility of scaling-up the electrolytic production of carbon nanotubes in molten salts has been in v estigated with the aid of electron microscopy (TEM and SEM). Using molten LiCl as the electrolyte and commercial graphite as both cathode and anodematerials, carbon nanomaterials, including nanotubes, were prepared by constant  v oltage electrolysis. The cell was more than 20times as large as that used in pre v ious work. The nanotube concentration in the final product increased with cell  v oltage (including iR  drop) from 1  v ol.% at 4.0 V to 35  v ol.% at 8.4 V. Under desired conditions, the charge and energy consumption for the cathodeerosion was 0.28 Ah/g and 4.1 Wh/g, of which 60    / 70 wt.% were for producing nanomaterials (nanotubes:   / 30  v ol.%). Whenadding 1 wt.% SnCl 2  to the electrolyte, partial and fully filled nanotubes were obtained with the nanomaterials containing up to 20wt.% Sn. Preliminary results from applying the product as the electrode in lithium ion batteries are reported. #  2002 Elsevier Science Ltd. All rights reserved. Keywords:  Carbon nanotubes; Molten salts; Electrolysis; Scaling up; Electron microscopy 1. Introduction Carbon nanotubes (CNTs) are strong, stiff, andelectrically conducti v e. They possess large aspect ratiosand surface areas, and can be either multi- or singlewalled structures [1]. Practical uses of CNTs havebecome more promising in recent years, such as fillersin polymer composites, field emitters for flat paneldisplays, electrodes for supercapacitors and elementsin nano-electronics [2    / 4]. CNTs are typically producedin the gas phase, by the e v aporation of pure carbonusing a high-powered energy source (e.g. electric arc,laser or solar heat) or by the catalytic decomposition of gaseous hydrocarbons o v er a transition metal (e.g.acetylene o v er Fe) [5    / 7]. Howe v er, in 1995, Hsu et al.disco v ered that CNTs could also be produced in moltenLiCl by electrolysis using high purity carbon electrodes,presenting the first example of producing CNTs in acondensed phase [8]. In their experiments, the carboncathode underwent erosion and nanometre sized pro-ducts, including multiwalled CNTs were found in themolten salt (electrolyte). Subsequently they conducted amore detailed study and concluded that for CNTproduction, the immersion of cathode in the electrolyteneeded to be shallow ( B / 3 cm), the current moderate( B / 5 A) and the  v oltage low ( B / 5 V) [9]. The disco v eryof Hsu et al. stimulated further work in this laboratoryunder  v arious conditions in different electrolytes (LiCl,NaCl and KCl) and, particularly, suitable CNT pre-paration temperatures for each electrolyte were estab-lished [10,11]. It was concluded that CNTs were not toform at temperatures higher than the boiling points of the alkali metals [10,11]. In 2000, Kaptay et al. reportedthat, on graphite cathode, the electro-deposition of Li,Na, K, Mg and Ca from the respecti v e molten chloridesall led to the formation of CNTs, but not of Sn and Ni[12].Hsu et al. [13] reported later that the electrolyticmethod could also be used to make carbon coated metalnanowires by adding a low melting temperature metal or * Corresponding authors. Tel.:   / 44-1223-762965; fax:   / 44-122-3334567 E-mail addresses:  gzc20@hermes.cam.ac.uk (G.Z. Chen),djf25@hermes.cam.ac.uk (D.J. Fray). 1 Academic Visitor from the Faculty of Technology andMetallurgy, Uni v ersity ‘St.Cyril and Methodius’, 16 RudjerBosko v ic, 1000 Skopje, R Macedonia.Electrochimica Acta 48 (2002) 91    / 102www.else v ier.com/locate/electacta0013-4686/02/$ - see front matter # 2002 Elsevier Science Ltd. All rights reserved.PII: S0013-4686(02)00595-9  salt in small concentrations to the electrolyte. Forexample, SnCl 2  added to LiCl resulted in the productionof nanotubes filled with Sn [13] and this is an ad v antageof the electrolytic method o v er some of the otherproduction methods since the filled nanotubes areproduced in situ.While all results from this and other laboratoriessuggest that the electrochemical formation of CNTs inmolten salts is related with the erosion of the cathode,such beha v iour of a carbon cathode is, as was firstpointed out by Fray [10,14], common in the Hall-Heroult cell where the intercalation of alkali metals,particularly sodium, into the carbon cathode is knownto ha v e caused cathode expansion, cracking and erosion.Based on the facts that CNTs were collectable from thesolidified salt phase and also the CNT yield dependedstrongly on electrolysis current and temperature [8    / 12],it was further proposed [11,15] that the alkali metal ion(M  ) intercalates, under the influence of a sufficientlynegati v e electrode potential, into the graphite latticewhere it is reduced in situ. The alkali metal then expandsthe lattice and more strain is put into the lattice as theamount of metal increases, until the lattice fragments.The carbon based fragments may then enter the moltensalt and, without the protection of the graphite lattice,undergo through an inter- and/or intra-fragment re-combination process, leading to the formation of  v arious carbon nanoparticles and nanotubes in theelectrolyte [11,12]. Alternati v ely, as it was initiallypostulated in this laboratory, the intercalation of alkalimetal could lead to the extrusion of carbon. Instead of forming CNTs in electrolyte  v ia the re-combination of carbon fragments, the outcome of extrusion may be thedirect formation of CNTs or their pre-cursors on thesurface of the cathode. Interestingly, this prediction hadnot been experimentally confirmed in this and otherlaboratories [8    / 14] until recently when Kaptay et al.reported their obser v ation of carbon micro-tubes on thesurface of a graphite cathode [16]. Ne v ertheless, the twoproposed post-intercalation processes can be consideredessentially the same if the extruded CNTs or their pre-cursors can fall into and grow further in the electrolyteor the inter- and/or intra-fragment re-combination mayoccur on the cathode’s surface.As reported in literature, the majority of researchersperformed constant current electrolysis in their prepara-tions of CNTs in molten salts and, e v en when  v oltagecontrol was used, results were mainly expressed in termsof current [9]. Howe v er, in constant current electrolysis,because of erosion, the surface area of cathode changeswith time and so does the current density. This isob v iously undesirable as the electrode reaction can besignificantly altered by current density  v ariation, mak-ing it difficult to correlate electrochemical parameterswith the products. In a two-electrode cell, a relati v elyconstant cathode potential, and hence the currentdensity, can be achie v ed if a constant cell  v oltage isapplied between the cathode and an anode of muchlarger surface area. Following this approach, constant v oltage electrolysis was carried out in molten NaCl inthis laboratory and the results exhibited a maximumCNT yield against cell  v oltages that was  v aried between3 and 9 V [15]. This obser v ation supports strongly theintercalation mechanism in which the speed of alkalimetal intercalation, which can be correlated with theelectrode potential (and hence the current density), playsa key role in determining the composition and structureof fragmented or extruded products [11,15].It should be pointed out that, up till now, all reportson the formation of CNTs in molten salts ha v edescribed electrolysis in relati v ely small cells. In addi-tion, the electrolysis times gi v en in these reports areusually short, which is due to firstly the completeerosion of the cathode and secondly the saturation of the salt by electrolysis products, particularly the alkalior alkaline earth metal. Therefore, it would be bothscientifically interesting and practically important toin v estigate the electrochemical method at increased v olume and time scales. In this work, a relati v ely largetwo-electrode cell, as shown in Figs. 1 and 2, was usedfor electrolysis. To compensate cathode erosion, step-by-step feeding of the cathode into the electrolyte wasalso applied during electrolysis. In addition, instead of current, the cell  v oltage was kept constant in eachexperiment and correlated with the CNT concentrationin the produced carbon nanomaterials that was esti-mated from electron microscopic analysis. The reportedresult of producing tin filled CNTs by adding SnCl 2  tothe electrolyte [13] was reproduced at an increased scale.Furthermore, the extraction of the produced carbonac-eous materials from the solidified salt phase wasin v estigated. Finally, the preliminary results of usingthe extracted carbonaceous product, containing about Fig. 1. A photograph graphite showing the crucible (left, large) used inthis work and that (right small) used pre v iously by Chen et al. [10,11].Each scale on the ruler on the far right is 1 cm. A.T. Dimitro v  et al. / Electrochimica Acta 48 (2002) 91    /  102 92  30  v ol.% CNTs, as the lithium intercalation electrode inlithium ion batteries are reported. 2. Experimental Electrolysis was carried out in a cylindrical graphitecell (external: 15.0 cm height and 9.0 cm diameter;internal: 14.0 cm height and 7.0 cm diameter, GraphiteTechnologies, high density grade) which also acted asthe anode during electrolysis. This cell had a  v olumesignificantly larger than those used pre v iously in thislaboratory [10,11] (see Fig. 1). A graphite rod (6.5 mm diameter, Graphite Technologies, EC4 grade, density:1.75 g/cm 3 ) was used as the cathode and was fed, step-by-step, into the electrolyte after electrolytic consump-tion. Molten LiCl or LiCl    / SnCl 2  (99:1 by weight, bothfrom Aldrich, ACS grade) was used as the electrolyte at700  8 C (Table 1). As was obser v ed pre v iously, theelectrolysis continuously produced alkali metal (Li)that could float on the surface of the molten salt,pro v iding a short circuit between anode and cathode. Topre v ent this, the cathode was surrounded with analumina sheath (external and internal diameters: 1.15and 0.8 cm) whose lower end was placed about 1.5 cmunderneath the surface of the molten salt. The cell andelectrode arrangement is illustrated in Fig. 2. The initialdepth of the molten salt was greater than 5.0 cm, and thecathode’s exposed length (outside the alumina sheath)was 1.0 cm, gi v ing rising to a surface area ratio of anode-to-cathode to be greater than 60.The graphite cell was placed in a sealable Inconel tubereactor held in a  v ertical 1600  8 C Lenton furnace. Theexperimental set-up was described in detail pre v iously[10,11] and is also schematically shown in Fig. 2. The salt was always thermally pre-dried inside the Inconeltube at 250  8 C, for at least 2 h and then melted under anargon flow (100 cm 3 /min). The electrolysis was con-trolled by a Farnell LS30-10 auto ranging Power supplyand carried out at constant  v oltages (4.0    / 8.4 V). Thecathode was inserted as the electrolysis progressed toreplace the cathode that had been eroded. The cathodeinsertion length was the same as the initial one and thetime of electrolysis was the same as the complete erosiontime that was determined separately at different  v ol-tages. The number of cathode insertions was up to 20,equi v alent to a total cathode length of 20 cm or 11.6 g of graphite being consumed (experiment 8, Table 1). Inaddition, due to the increase of cell  v olume, the amountof the electrolyte (LiCl) used in this work was up to 250g, which is more than 20 times as much as that used inpre v ious work [11]. Howe v er, it should be noted that,because the cell also functioned as the anode, the use of a larger cell increased the cathode-to-anode distancefrom less than 1.0 cm in pre v ious work [10,11] to o v er3.0 cm, leading to greater  iR  drops across the electrolyte( v oltage losses due to the electrolyte resistance).After ha v ing progressed for a predetermined period,counted by the number of cathode insertions, theelectrolysis was terminated. The cathode was thenremo v ed from the reactor and the electrolyte left tosolidify o v ernight under a slow flow of argon ( B / 10 cm 3 /min). The carbonaceous materials produced were har- v ested, in principle, using the same methodology aspre v iously reported [8    / 11], i.e. (i) dissol v ing the solidi-fied electrolyte in water; and (ii) extracting the carbo-naceous materials by adding toluene to the aqueousmixture, resulting in the carbonaceous materials aggre-gating at the water    / toluene interface. In this work,howe v er, additional washing steps were applied betweenthe dissolution and toluene extraction, aiming at remo v- ing the electrolyte and related impurities that might ha v ebeen trapped in the carbonaceous materials.Complete dissolution of the carbon containing soli-dified electrolyte (100    / 250 g) was achie v ed within anhour by one or two additions of 250    / 300 ml water toand mechanical stirring inside the cell (the solubility of LiCl in water is 81 and 107 g at 25 and 80  8 C,respecti v ely). The black aqueous mixture in the cellwas poured into a large beaker to which more water(500    / 2000 ml) was added. The solution was furtherstirred for another hour, and then left to stand stillo v ernight, gi v ing time for the carbonaceous material tofully precipitate. Afterwards, the clear top solution wasdecanted. This washing procedure might be repeatedonce or twice. To impro v e the washing efficiency, insome cases, hot water (80  8 C) was used and mechanical Fig. 2. A schematic diagram of the experimental set-up for molten saltelectrolysis. A.T. Dimitro v  et al. / Electrochimica Acta 48 (2002) 91    /  102  93  stirring was carried out while placing the beaker in anultrasonic bath.After washing and decanting most of the water, asmall amount of the carbonaceous material was driedand analysed directly and the rest was transferred into alarge separation flask (1.0 l) containing mixed water andtoluene (10:1,  v ol.), followed by thorough shaking. Theflask was then allowed to stand still for phase separa-tion. As reported pre v iously, after phase separation, thecarbonaceous materials were found to aggregate at thewater/toluene and toluene/glass wall interfaces [10].Once separated, the water phase was drained. Thecarbonaceous materials and toluene were transferredinto a storage bottle (with lid). The toluene could be re-collected by the con v entional e v aporation method, andre-used in the next extraction. Howe v er, for mostsamples, some toluene was kept in the storage bottlewith the carbonaceous materials. This was because itwas much easier to simply shake the bottle to tempora-rily and uniformly suspend the carbonaceous materialsand take a small amount of the suspension for furtheranalysis, e.g. by electron microscopy which was per-formed on JEOL 200CX (TEM, 200 kV) and JEOL6340F (SEM, 5 to 20 kV, capable of energy dispersi v eX-ray analysis). The CNT concentration in the collectedcarbonaceous materials was estimated from SEMimages of relati v ely low magnifications [8    / 11].Some of the samples were sent to AEA Technologiesand made into the lithium intercalation electrodes whosecapacity and rechargeablity were examined in a testinglithium ion battery by the standard method de v elopedby AEA Technologies. 3. Results and discussion 3.1. Electrolysis In pre v ious work, electrolytic preparation of CNTs inmolten salts was conducted using naked graphite rodcathodes and constant current [8    / 13,16]. Consequently,due to its lower density than the corresponding moltenchloride salt, the electrolytically formed alkali or alka-line earth metal floated on the electrolyte’s surface andpro v ided a short circuit between the anode and cathode.While such a problem might not be  v ery serious forbatch preparation, it will certainly pre v ent further orcontinuous operations from using the same molten salt.In this work, in order to a v oid the floating alkali metal(Li) shorting the electrodes, it was decided to use analumina sheath to surround the upper portion of thecathode in contact with the electrolyte, as shown in Fig.2. In addition, as discussed in Section 1, constant v oltage electrolysis was applied. Because such experi-mental arrangements ha v e ne v er been applied in pre- v ious work, it is worth reporting firstly the cellperformance in the batch operation mode.In this work, all electrolysis experiments were con-ducted at 700  8 C which is in the optimal temperaturerange for CNT preparation in molten LiCl as deter-mined in pre v ious work [10,11]. Fig. 3 plots the currents measured at different times and  v oltages when electro-lysing a fresh bath of pure molten LiCl for the firstcathode insertion (Fig. 3a), and that at 7.6 V for anumber of consecuti v e insertions (Fig. 3b). For the firstinsertion (Fig. 3a), apart from the strong dependence of the current on the applied cell  v oltage, the plots showclearly two processes: the current drops relati v elyrapidly in a short period soon after applying the  v oltageand then declines slowly with time, either continuouslyat low  v oltages (Fig. 3a) or passing through one or morepeaks at high  v oltages (Fig. 3b). It is worth mentioningthat, when a high  v oltage (  / 7.0 V) was applied,intermittent cracking noises were heard from the reactorduring electrolysis. Simultaneously, the current was seento rise and fall on the display of the power supply. Insome cases, the rise and fall of the current were as sharpas a spike (not shown in Fig. 3 which were recordedmanually) and, in others, took a relati v ely longer time,forming the broad current peaks as shown in Fig. 3b. On Table 1Experimental conditionsExperimentnumberElectrolyte a Cell  v oltage(V)Initial current density(A/cm 2 )Number of 1.0 cminsertionsCNT concentration in product 9 5  v ol.%Composition Mass (g)1 LiCl 100 4.0 1.0 6 12 LiCl 100 5.2 1.5 6 53 LiCl 100 6.5 2.0 6 104 LiCl 100 7.6 2.5 6 255 LiCl 100 8.4 3.0 6 356 LiCl  1%SnCl 2  100 8.4 3.0 6 257 LiCl 250 8.4 3.0 15 358 LiCl  1%SnCl 2  250 8.4 3.0 20 25 a Electrolysis was carried out at 700  8 C under argon (100 cm 3 /min). A.T. Dimitro v  et al. / Electrochimica Acta 48 (2002) 91    /  102 94  the basis of a pre v ious report of a cracked graphitecathode resulted from electrolysis in NaCl [11], it isbelie v ed that the noises came from the cracking of thecathode caused by fast intercalation and in situ reduc-tion of Li  ion. Apparently, if the cracking was smalland quick, leading to the detachment of broken pieces, acurrent spike could form; if the cracking was large andslow, allowing the broken pieces to hang onto thecathode for a moment before falling into the electrolyte,the effect would then be a temporary increase in thesurface area of the cathode and hence the relati v ely slowcurrent rise and fall.When terminating the electrolysis at the end of theinitial current drop and remo v ing the cathode from thereactor, it was obser v ed that most of the cathodeunderneath the alumina sheath disappeared, particularlyin high  v oltage experiments. After the initial fast erosionstage, howe v er, it took a much longer time for the smallremaining portion of the cathode to erode completely,which is reflected by the slowly declining section of thecurrent    / time plots in Fig. 3a. This phenomenon may befurther explained as follow. The decrease of current withtime is generally associated with the decrease of thecathode’s surface area due to erosion. According to theintercalation mechanism, the erosion results from thefast intercalation of Li  ion into and its reduction insidethe graphite lattice [11]. What has not been mentioned inthe Li  ion intercalation mechanism is the highly likelysimultaneous formation of liquid Li metal at thecathode. The formed liquid Li metal may ha v e twonegati v e effects on the intercalation process. It reducesthe current efficiency for the erosion of the cathode andmore importantly, if accumulated quickly, co v ers up thesurface of the cathode, pre v enting further intercalationof Li  ion. At the early stage of electrolysis, the formedLi metal can be remo v ed from the cathode surface  v iadissolution into the molten salt. Also, the erosion of thecathode, particularly  v ia cracking, not only disturbs theaccumulation of Li metal, but also allows the renewal of the cathode’s surface. Howe v er, when the electrolytenear the cathode becomes saturated, Li metal accumu-lates and gradually co v ers up the cathode’s surface. Thefact that, in this work, the remaining portion of thecathode had a surface facing downward could ha v e alsohelped to entrap the Li metal. The gradual co v er up of the cathode surface by Li metal slows the o v erallintercalation of Li  ion and hence the speed of erosion,leading to the slowly declining current.The increase of current with increasing cell  v oltage,see Fig. 3a, is not unusual. Howe v er, the explanationmay be slightly more complicated, considering all thepossible processes at the cathode. The first contributioncomes, as expected, from the potential determined speedof electron transfer for the reduction of Li  ion to themetal, either on the surface or within the lattice of thegraphite cathode. Secondly, it has been theoretically andexperimentally confirmed, in a separate programme of this laboratory focusing on the mechanism of lithiumintercalation into graphite [17], that the speed of Li  ion intercalation also increases with the cathodic shift of the electrode potential. Thirdly, the formed Li metal,either dissol v ed in or floated on the surface of theelectrolyte, cannot only undergo re-oxidation at theanode but also pro v ide paths for electronic conduction,both shall reduce the current efficiency. Apparently, interms of impro v ing current efficiency for CNT prepara-tion, the formation of Li metal on the surface of thecathode is a negati v e factor but unfortunately ine v itabledue to the  v ery negati v e potential required to enable fastLi  ion intercalation. 3.2. Feeding cathode For scaling up the electrolytic production of CNTs, inaddition to an enlarged reactor (electrolytic cell), itwould be more desirable if the production may beconducted continuously. This should then include con-tinuous feed of reactants into and remo v al of productsfrom the reactor. In this work, the reactants were thecathode and molten LiCl and the products includedcarbonaceous materials and liquid lithium metal from Fig. 3. Current    / time plots of electrolysis in molten LiCl at 700  8 Cunder argon using a graphite rod cathode and a graphite crucibleanode: (a) at different constant  v oltages for the first cathode insertionin fresh electrolyte; and (b) at 7.6 V for consecuti v e insertions in thesame electrolyte. Cathode diameter, 0.65 cm; insertion depth, 1.0 cm;anode surface area, 150 cm 2 ; anode-to-cathode separation,   / 3.2 cm. A.T. Dimitro v  et al. / Electrochimica Acta 48 (2002) 91    /  102  95
Related Search
Similar documents
View more...
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks